Supplementary Figures - Word file (622 KB )

advertisement
Normalized EPSP slope (%)
200
Proximal
200
100
100
50
50
-200 -150 -100 -50
0
50 100
Distal
-200 -150 -100 -50
0
50 100
Pre/post interval (ms)
Supplementary Figure 1. Spike-timing-dependent synaptic depression at proximal and distal
dendrites in the presence of 10 M picrotoxin (black circles), superimposed on the STDP window
measured in normal bath solution (gray circles, same as Fig. 1d). At 100<t<50 ms, significant
LTD was induced at distal (–25.2±6.1%, n=6, p<0.01) but not at proximal dendrite (6.8±5.6%, n=4,
p>0.8).
a
c
4AP
Control
150
100
50
0
0
50
100
50
25
50
100
Spike amplitude (mV)
4
0
0
Suppr. width (ms)
75
0
8
150
50
100
150
Distance (m)
e
100
0
12
150
Distance (m)
d
105m
Spike width (ms)
Spike amplitude (mV)
b
Suppr. amplitude (%)
75m
35m
60
40
20
0
0
4
8
12
Spike width (ms)
Supplementary Figure 2. Back-propagating APs and AP-induced suppression of EPSPs recorded
from apical dendrites of L2/3 pyramidal cells. a, Examples of back-propagating APs recorded at
35, 75, and 105 m from the soma (from different cells). APs were evoked antidromically with a
small stimulating electrode placed at the axon initial segment. Scale bars: 40 mV, 5 ms. b, Spike
amplitude as a function of dendritic distance from the soma. Red symbols indicate experiments
performed with 4-AP (3 mM) in the recording electrode. c, Spike width (measured at half-height of
the AP) as a function of dendritic distance. d, Amplitude of EPSP suppression window as a
function of spike amplitude. Each suppression window was fit with a single exponential 1  Ae  t /  ;
A is used to represent amplitude. e, Width of EPSP suppression window (measured at half-height of
the exponential fit, which equals  log 2 ) as a function of spike width.
Normalized amplitude (%)
Dendritic
AMPAR-EPSPs
Dendritic
NMDAR-EPSPs
120
120
100
100
80
80
40
dist
dist
prox
60
-150
-100
prox
60
-50
Pre/post interval (ms)
0
40
-150
-100
-50
0
Pre/post interval (ms)
Supplementary Figure 3. Suppression of AMPAR-EPSP and NMDAR-EPSP by back-propagating
AP. Same as Fig. 2e, f, except that the recordings were made in the dendrite near the stimulating
electrode. n=2 to 6 (AMPAR-EPSPs) and 3 to 7 (NMDAR-EPSPs).
b
NMDAR-EPSP
suppression (dist)
120
100
80
Control
NR2A871-950
60
40
-150
-100
-50
Pre/post interval (ms)
0
Synaptic modification (%)
Normalized amplitude (%)
a
LTD
150
100
50
0
Control NR2A871-950
Supplementary Figure 4. Expression of NR2A871-950 blocks NMDAR-EPSP suppression and LTD
induction. a, Temporal windows of AP-induced suppression of NMDAR-EPSPs at distal dendrite
of infected (black line, n=4) and control (gray line, n=7 to 12) L2/3 pyramidal neurons. Error bar: 
s.e.m. Data were significantly different over the following pre/post intervals: –10 ms (p<0.05), –20
ms (p<0.01), –30 ms (p<0.001), –50 ms (p<106), –75 ms (p<0.01), and –100 ms (p<0.05); t test. b,
LTD was induced by 60 postpre spike pairs for control (–40.4±4.3%; n=18) but not for infected
neurons (–5.1±3.9%; n=3; p>0.4, t test). The difference between control and infected cells was
significant (p<0.001, t test).
Methods. Using standard molecular biology techniques, GFP was inserted before a sequence
corresponding to residues 871-950 of the NR2A subunit of the NMDAR, a serine-containing region
of the C-terminus important for calcineurin-dependent NMDAR desensitization (Krupp et al., 2002,
Neuropharmacology 42:593). Semliki Forest Virus (SFV) containing this construct was made by
co-transfecting HEK293 cells with pSCA3-GFP-NR2A and pSCAhelper using lipofectamine 2000.
After 48 hours, culture media was harvested and virus was activated with 0.5mg/ml -chymotrypsin
(45 min, room temperature). Surgeries were performed in 4-week old rats in accordance with the
guidelines of the Animal Care and Use Committee at the University of California, Berkeley. Rats
were anesthetized with ketamine (80 mg/kg) and xylazine (10 mg/kg). Small craniotomies (1 mm 2)
were performed over the visual cortex and SFV-containing media was injected into the superficial
cortex with a small pipette (15 μm tip diameter). Rats were allowed to recover for two days before
brain slices were cut (Takahashi et al., 2003, Science 299:1585). Infected cells were identified
based on GFP labeling. Resting membrane potentials and input resistance of infected neurons were
not significantly different from age-matched control neurons (p>0.3).
Supplementary Figure 5. Dendritic Ca2+ signal evoked by back-propagating AP. a, Two-photon
fluorescence image of a layer 2/3 pyramidal neuron filled with 200 M Oregon Green BAPTA 1.
Scale bar: 20 m. b, Upper traces: Fluorescence Ca2+ signals at distal (108 m) and proximal (22
m) dendrites (white boxes in a) in response to single back-propagating APs evoked by somatic
current injection, each averaged from 11 to 18 trials. Lower plot: distal vs. proximal Ca2+ signals
evoked by single APs, measured by either peak amplitude () or peak area (integral, , in arbitrary
units). Each symbol represents data from one cell (n=10); open symbol: with 100 M Oregon
Green BAPTA 1; filled symbol: 200 M. c, 4-AP enhances the AP-induced Ca2+ signal at proximal
dendrite. Upper traces: Ca2+ signals at proximal dendrite of the cell shown in (a) in the presence of
extracellular 4-AP and after washout. Lower plot: AP-induced Ca2+ signals at proximal dendrite
after vs. before bath application of 4-AP (n=6).
Method. Oregon Green BAPTA 1 (100-200 M, Molecular Probes) was loaded into the cell via the
somatic whole-cell pipette (loading time before imaging: ~15 min). Imaging was done using a twophoton laser scanning system consisting of an upright microscope (DM LFS, Leica Microsystems,
Heidelberg) with 40×W0.8IR objective and a “Tsunami” titanium-sapphire laser pumped by a
“Millennia” neodymium yttrium vanadate solid state laser (SpectraPhysics, Mountain View, CA),
which provides <100 fs pulses at 890 nm at 80 MHz. Line-scan mode was used to record single
AP-induced Ca2+ transients (temporal resolution: 1 ms). At each dendritic location, 5 to 71 sweeps
were measured (interleaved between distal and proximal sites), and results were averaged. Change
in Ca2+ concentration is given in the change of fluorescence normalized by baseline fluorescence
(ΔF/F) after background correction. Data were analyzed in Matlab (MathWorks).
a
-
b
30 ms
15 ms
7.5 ms
3 ms
Probability
Stimulus
+
Presynaptic
spike trains
9 ms
0
18 ms
c
50
67.5 ms
Postsynaptic
spike train
Synaptic
modification (%)
37.5 ms
0
-50
50 ms
-100
0
100
Pre/post interval (ms)
d
...
...
...
...
...
...
...
...
Synaptic
weight
Before training
1
0
0
1
15
30 45
 (ms)
60 75
0
0
1
15
30 45
 (ms)
60
75
0
0
1
15
30 45
 (ms)
60
75
0
0
15
30 45
60
75
30 45
60
75
 (ms)
Synaptic
weight
After training
1
0
0
1
15
30 45
 (ms)
60 75
0
0
1
15
30 45
 (ms)
60
75
0
0
1
15
30 45
 (ms)
60
75
0
0
15
 (ms)
Supplementary Figure 6. Further analyses of the model on location-dependent input selection. a,
Sensory stimulus used in simulation and spike trains of example model neurons. Number on the
right of each presynaptic spike train indicates time constant of impulse response function of the cell
(isee Supplementary Methods). To effectively illustrate the relative spike timing, spike trains are
shown for a selected period with high-frequency bursts from many presynaptic neurons; the overall
mean firing rate of the cells were much lower. b, Cross-correlogram between pre- and postsynaptic
spike trains. Warmer colors represent inputs with more transient responses (numbers on the right
indicate i). The number of pre/post spike pairs contributing to LTP and LTD can be roughly
estimated from the dark and light shaded areas under each curve; dark shading: width of LTP
window in c (+=12.5 ms), light shading: width of LTD window (=103.4 ms). c, Single
exponential fit of the STDP window measured at distal dendrites. d, Simulation results of a 4compartment model. Shown are diagram of synaptic inputs (upper), and synaptic weight vs. i
before (middle) and after (lower) training. The location dependence of input selection is similar to
that found in the 2-compartment model (Fig. 5b).
Download